A FTERMARKET P ERFORMANCE

In the preceding sections, I outline the existing literature on IPOs when it comes to underpricing and cyclicality. In this section, I describe existing evidence of aftermarket performance of IPOs, focused on the UK. Unlike underpricing, IPOs tend to exhibit negative aftermarket performance in the long-run.

In his study, Levis (1993) uses three alternative benchmarks to measure aftermarket performance of UK IPOs in the 1980s. The three benchmarks used are the market capitalization weighted FTA index, the capitalization weighted HGSC index, and the all share equally weighted index. Based on these benchmarks, he finds a 36-month return excluding the first month returns equal to -11.4%, -8.3%, and -23.0% for the FTA, HGSC, and all share index, respectively. Goergen and Renneboog (2003) find that over a period of five years from a total of 764 British firms that went public between 1981 and 1988, that the IPOs have an aftermarket performance of -33%. Furthermore, Espenlaub, Gregory et al. (2000) find evidence that after 36 months, UK IPOs issued between 1985 and 1992 underperform based on the CAPM and Fama-French three-factor model, -15.9% and -28.2%, respectively.

However, the 60 month returns shows that the deterioration of returns slows down, depending on the benchmark chosen, with cumulative abnormal returns ranging from -4.3% to -42.8%

(Espenlaub, Gregory et al. 2000). Looking at PE- and VC-backed IPOs, Levis (2011) compares these two groups to a selection of non-sponsored UK IPOs from 1992 to 2005. Using various FTSE benchmarks, he shows that based on the Financial Times All-Share Index, PE-backed IPOs have a positive 36 month equally-weighted return of 13.8%, while VC-PE-backed and non-sponsored IPOs both had negative returns of -3.9% and -20.2%, respectively. Overall, the sample of IPOs generate an abnormal return of -13.5% after 36 months, demonstrating that

18

overall IPOs underperform in the long-run. However, both Levis (2011) and Espenlaub, Gregory et al. (2000) find that the amount of underperformance depends on the benchmark chosen. For my sample of 194 UK IPOs between 2006 and 2017, I find varying degrees of both underperformance and outperformance. The six-month holding period exhibits outperformance relative to the FTSE All Share Index and FTSE Small Cap Index equal to 4.9% and 4.4%, respectively, on a cumulative abnormal return basis. The six-month period is not often analysed, but I chose to analyse it because of the common lockup period of 180 days.

For the five-year buy-and-hold abnormal returns, the UK exhibits underperformance to the FTSE All Share Index and FTSE Small Cap Index equal to -7.0% and -9.3%, respectively.

Having examined the literature surrounding aftermarket performance, I review what drives the performance next.

Ritter and Welch (2002) favour the behavioural point of view when observing long-term aftermarket underperformance. They find an average marketadjusted underperformance of -23.4% over three years for US firms between 1980 and 2001 (Ritter and Welch 2002).

However, this is highly dependent on the benchmark chosen. Furthermore, Espenlaub, Gregory et al. (2000) finds statistically significant negative aftermarket performance regardless of the benchmark used. They find that the oil and gas industry consistently performs the worst, along with the fact that smaller firms tend to experience larger underperformance (Espenlaub, Gregory et al. 2000). In my study, I find that the aftermarket performance varies depending on the benchmark chosen, and I therefore focus my study on two benchmarks, the FTSE All Share Index and FTSE Small Cap Index, in order to get a picture based on the overall market and for the smaller firms. However, I do not uncover many significant differences between industries.

From a timing perspective, Loughran, Ritter et al. (1994) find that there is a tendency for high volume timeframes to experience lower long-term returns. Specifically in the US, older and more established firms have smaller initial average returns, but higher long-term returns when compared with their younger counterparts (Loughran, Ritter et al. 1994). Although as mentioned above in Section 2.2, I find differences among years in aftermarket performance, I do not see significant differences in periods defined as hot or cold, as defined by IPO volume or IPO initial returns.

19

Ritter (1991) attributes long-term aftermarket underperformance to risk mismeasurement, bad luck, or fads and over optimism. To explain whether risk mismeasurement plays a role, he uses different benchmarks to find robust results (Ritter 1991). Levis (1993) takes it a step further by finding that firms with the highest initial return tend to experience the worst aftermarket performance in the long-term⁴. Inspired by Levis (1993), I find that the IPOs that have initial returns equal to or greater than the median in the sample experience a boost on the six-month buy-and-hold abnormal returns ranging from 9.4% to 11.9% depending on the benchmark and inclusion of first month return. However, only the three-year cumulative abnormal returns are significant and equal to a 28.5% boost on the average three-year return, which is contrary to Levis (1993)

Finally, academic literature related to and VC-backed IPOs, Levis (2011) finds that PE-backed IPOs consistently have positive and significant cumulative abnormal returns which is related to the firms’ leverage ratios and proportion of shareholders, on an equally- and value-weighted basis. On the contrary, VC-backed IPOs and other non-sponsored IPOs perform poorly in the aftermarket (Levis 2011). My results are mixed, where PE-backed IPOs see a boost in the high teens for the six-month buy-and-hold returns and this approximately doubles to 30% when looking at the one-year holding period. This is less pronounced in the longer holding periods of three- and five-years. Furthermore, I find no significant underperformance for VC-backed IPOs, although the coefficients are negative.

Having discussed the literature surrounding IPOs related to underpricing and aftermarket performance, the next section discusses the data collection process and the characteristics of the data.

4 Over a period of 36 months.

20